Essential Role of the Unusual DNA-binding Motif of BAG-1 for Inhibition of the Glucocorticoid Receptor* 210

The co-chaperone BAG-1 is involved in the regulation of steroid hormone receptors, including the glucocorticoid receptor (GR). More recently, BAG-1 was found in the nucleus where it decreases GR transactivation. Moreover, nonspecific DNA binding of BAG-1 has been reported. We discovered that of the N-terminal part of BAG-1M, the first 8 amino acids are sufficient for DNA binding, containing a stretch of three lysines and a stretch of three arginines. Changing the spacing between these stretches had no effect on DNA binding. Surprisingly, this small, nonsequence-specific DNA binding domain was nonetheless necessary for the inhibitory function of BAG-1 for GR-dependent transcription, whereas the following serine- and threonine-rich E2 X 4 repeat domain was not. Mutational analysis of these two domains revealed that only mutants retaining DNA binding capability were able to down-regulate GR-mediated transactivation. Intriguingly, lack of DNA binding could not be functionally rescued by BAG-1M harboring a point mutation abolishing interaction with hsp70. Thus, DNA binding and hsp70 interaction are required in cis. We propose that the nonsequence-specific DNA-binding protein BAG-1 acts at specific chromosomal loci by interacting with other proteins.

BAG-1 is involved in the regulation of a diversity of physiological processes, e.g. apoptosis, tumorigenesis, and neuronal differentiation (1)(2)(3)(4), by virtue of its ability to interact with numerous regulatory proteins, including nuclear receptors. BAG-1 was originally identified as an associating factor of the anti-apoptotic factor bcl2 (5) and, independently, as a protein called RAP46 that associates with the glucocorticoid receptor (GR) 1 and other steroid hormone receptors (6). It also is termed "hap46," hsp70-and hsc70-associating protein (7). BAG-1 enhances the transcriptional activity of the androgen receptor (8,9), and in the case of the vitamin D receptor evidence has been provided for both enhancement (10) and inhibition (11) of tran-scriptional activity. In addition, BAG-1 interaction with the retinoic acid receptor (RAR) results in inhibition of binding of RAR to its response elements on DNA and of RAR-dependent transcription (12). Likewise, BAG-1 has been described as a negative regulator of GR, because it binds to the hinge region of GR and inhibits DNA binding and transactivation of the receptor (13).
Before hormone activation, GR resides in the cytoplasm, where it interacts with various molecular chaperones in a stepwise fashion to attain the state competent for hormone binding (14 -16). Central to this folding process are heat shock protein (hsp) 90, hsp70, and hsp70/hsp90 organizing protein (hop) (17)(18)(19), which bridges hsp70 and hsp90 via its tetratricopeptide repeat domains (20). hsp90, hsp70, and hop, possibly with the involvement of hsp40 (21), form an intermediate complex with GR (15,22), from which hsp70 and hop presumably dissociate to allow entry of p23 and one of the large immunophilins to the final complex, where GR gains hormone binding competence.
The chaperone activity of hsp70 can be modulated by BAG-1 and also by hsp40, C terminus of hsp70-interacting protein, and hsp70-interacting protein (hip). hsp40 enhances the ATPase activity of hsp70 in vitro (23) and the hsp70-dependent refolding in mammalian cells (24,25). hip also has been identified as a positive regulator of hsp70 chaperone activity (26,27). In contrast, the C terminus of hsp70-interacting protein has been found to inhibit the ATPase activity of hsp70 and to interfere with stable hsp70-substrate complexes (28). All isoforms of BAG-1, i.e. the 50-kDa BAG-1L, the 46-kDa BAG-1M, and the 33/29-kDa BAG-1S, which derive from different translation initiation sites localized on the same gene (29 -31), have been described in several studies (27,(32)(33)(34)(35) to inhibit hsp70dependent refolding activity in vitro and in vivo. Moreover, BAG-1 was found to compete with the stimulatory action of hip (27). hip, in turn, opposes the negative effect of BAG-1 on steroid binding of GR (36). The negative effect of BAG-1 on steroid binding (37) was not observed by others (38). This seeming discrepancy could be explained by different protein concentrations of BAG-1 present in these assays, as the protein level of BAG-1 has been shown to be crucial for its inhibitory action (37).
Although all these reports strongly suggest a cytosolic effect of BAG-1 on GR folding and activity, there are also data supporting a nuclear function of BAG-1. For example, BAG-1 has been reported to bind to DNA in a nonsequence-specific manner and to stimulate DNA transcription (39). Deletion of, or mutations within, the N-terminal 10 amino acids of BAG-1 abolish its DNA binding (39). In addition, it has been shown that BAG-1 is transported into the nucleus upon steroid binding and nuclear translocation of GR (38). This nuclear translocation is * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. dependent on the C-terminal hsp70 binding domain of BAG-1. Moreover, the inhibiting effect of BAG-1 on DNA binding of GR (13) can be overcome in a cell-free system by supplementing with increasing amounts of hsp70 (40).
Both C-and N-terminal deletions deprive BAG-1 of its ability to inhibit GR function (38). Whereas an hsp70 interaction domain has been identified and characterized by crystallography and NMR analysis in the C-terminal region (41,42), the function of the N-terminal part seems less clear. It has been speculated that a serine-and threonine-rich E 2 X 4 domain may be necessary for the inhibitory function of BAG-1 (38). However, it also is conceivable that the described DNA binding domain is important for inhibition of GR function.
Here we present a detailed mutational analysis of the Nterminal part of BAG-1 with the aim to 1) describe the unusual DNA binding domain in more detail, and 2) dissect the contributions of the E 2 X 4 motif domain and the DNA binding domain to the inhibition of GR-dependent transcription. We demonstrate here that, surprisingly, the first eight amino acids of BAG-1 are required for the inhibitory effect of BAG-1 on GR, whereas the E 2 X 4 domain is dispensable. Moreover, the binding of BAG-1 to DNA is because of the positive electric charge at the N terminus. Mutations that inhibit binding to DNA cannot be functionally rescued by overexpressing BAG-1 with a point mutation abolishing its interaction with hsp70, indicating that the DNA binding and the hsp70 interaction domain must be present in cis. Here we discuss a model in which the nonspecific DNA-binding motif of BAG-1 is essential for inhibiting GR function and becomes specific due to interaction with other factors like GR.
Cells were harvested at about 70 -90% confluency, and about 0.5 to 1 ϫ 10 7 cells were resuspended in 400 l of electroporation buffer (50 mM K 2 HPO 4 , 20 mM KAc, pH 7.35, 25 mM MgSO 4 ). 3.5 g of steroidresponsive firefly luciferase reporter plasmid MTV-Luc (44), 4 g of pRK7GR that expresses human GR (45) from the cytomegalovirus promoter of the vector pRK7 (46) or an HA-tagged version of it, 6 g of either one of the BAG-1 expression plasmids (Tables I and II) or the corresponding empty expression vector (to keep the total amount of transfected DNA constant), and as internal control plasmids either 3.5 g of simian virus 40 (SV40) promoter-driven ␤-galactosidase expression vector pCH110 (Amersham Biosciences) or 1 g of SV40-driven Renilla luciferase-driven expression vector (Promega) were added. Transfection was performed using an electroporation system (Biotechnologies & Experimental Research, San Diego) after determination of the optimal electrical field strength (47). Electroporated cells were replated and cultured for 24 h in fresh medium (containing 10% steroidfree FCS) supplemented with 100 nM dexamethasone (Sigma) or the solvent of dexamethasone (i.e. ethanol).
BAG-1 Expression Plasmids-A series of BAG-1 mutants was generated by PCR and cloned into the mammalian expression plasmid pRK5mcs. The plasmid pRK5mcs was derived from pRK5SV40PUR (46) by cutting with the restriction endonucleases EcoRI and HindIII and inserting an annealed linker oligonucleotide with the sequences 5Ј-AATTCTCGAGATATCGGGCCCGGATCCGCGGCCGCTCGCGA-3Ј and 5Ј-AGCTTCGCGAGCGGCCGCGGATCCGGGCCCGATATCTCG-AG-3Ј for the opposite strand. For cloning, the wild type and mutant sequences of BAG-1 were amplified by PCR from the plasmid pMal-c2-Bag-1 (34). We chose the vector pProExHTa (Invitrogen) to express BAG-1 proteins in bacteria. Sequences of the oligonucleotides used are provided as Supplemental Material.
Luciferase and ␤-Galactosidase Assays-Firefly luciferase and ␤-galactosidase assays were as described before (48). Either cells were scraped in 200 l of lysis buffer (0.1 M KHPO 4 , pH 7.8, 1 mM dithiothreitol) and cytosolic extracts were made by three freeze and thaw cycles and subsequent centrifugation. 50 l of each supernatant (corresponding to ϳ1-2 ϫ 10 5 cells) were transferred to a 96-well plate. 150 l of 33 mM KHPO 4 , pH 7.8, 1.7 mM ATP, 3.3 mM MgCl 2 , 13 mM luciferin (Roche Molecular Biochemicals) was added to each sample by the injector of an automatic luminometer (Luminat LB 96, Wallac GmbH, Freiburg, Germany), and light emission was measured for 10 s. To correct for variations in transfection efficiencies, values of the luciferase assay were normalized to either ␤-galactosidase or Renilla luciferase activities. To measure ␤-galactosidase activity, 50 l of cell extract was added to 100 l of galactosidase buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgCl 2 , 50 mM ␤-mercaptoethanol) on a 96-well plate. 20 l of 2 mg/ml O-nitrophenyl ␤-D-galactopyranoside was added, and the reaction was incubated at 37°C. After 10 -30 min, absorption was measured at 405 nm in a multiphotometer (Dynatech MR5000).
When using the Renilla luciferase expression plasmid in combination with the firefly reporter plasmid, cell extracts were scraped in 200 l of a lysis buffer provided and prepared according to the manufacturer's recommendation (Promega Inc.), and firefly and Renilla luciferase activities were measured on a 96-well plate in 50 l of cell extract in the automatic luminometer on a 96-well plate.
Protein Expression and Purification-Histidine-tagged BAG-1 encoding plasmids were grown in pBL3Lys bacteria (Invitrogen). Lysis and purification was performed using Ni-NTA-agarose according to the manufacturer's recommendations (Qiagen). The histidine tag was removed by cleavage with tobacco edge virus protease for 6 h at 30°C (Invitrogen). The cleaved proteins were rebound to Ni-NTA-agarose columns again to remove traces of uncleaved proteins and purified afterward with Bio-Spin columns (Bio-Rad) to remove dithiothreitol from cleavage buffer and traces of liquid columns. Protein concentrations were determined using the BCA Protein Assay kit (Pierce).
DNA Binding and Gel Shift Assays-DNA binding of BAG-1 was essentially performed as described (39). Briefly, 1 or 2 g of purified and tobacco edge virus-cleaved BAG proteins (isoforms or mutant) were incubated with 0.2 ng of 32 P-end-labeled 125-bp /HindIII DNA fragment in binding buffer (39) for 30 min at room temperature. Protein DNA complexes were resolved on native 5% acrylamide gels in TBE buffer and visualized after overnight incubation on x-ray films (Eastman Kodak).
Immunodetection of GR and BAG-1 Proteins-25 g of total protein from whole cell lysates used for luciferase assays was separated by SDS-PAGE under denaturing conditions (Lä mmli buffer containing 10% SDS). The proteins were transferred to polyvinylidene difluoride membrane (Schleicher & Schü ll). Nonspecific binding to the membrane was blocked by 5% nonfat milk in Tris-buffered saline/Tween buffer, and then specific antibodies were added. BAG protein was detected by addition of a monoclonal anti-BAG antibody (anti-BAG-C16, Santa Cruz Biotechnology, Inc., Santa Cruz, CA) followed by horseradish peroxidase-conjugated goat anti-rabbit antibody (Sigma). The HA tag on GR was detected by a monoclonal anti-HA antibody (Boehringer Ingelheim) followed by addition of horseradish peroxidase-conjugated goat anti-rat antibody (Sigma). Signals were visualized by enhanced chemiluminescence system solution (Amersham Biosciences).

Deletion of the N-terminal DNA Binding Domain of BAG-1 Abolishes Its Inhibitory Function on GR, Whereas Deletion of the E 2 X 4 Motif Domain Does Not-BAG-1S
, the short isoform of BAG-1, is unable to inhibit the transcriptional activity of GR (38). The N-terminal amino acid stretch missing in BAG-1S as compared with BAG-1M contains a serine-and threonine-rich E 2 X 4 repeat domain and a recently described DNA binding domain (Table I) (39). To clarify which features in the N terminus are required for inhibition of GR, we created two deletion mutants, either missing the E 2 X 4 domain (BAG-1M ⌬11-67, Table I) or the putative DNA binding domain (BAG-1M ⌬N10, Fig. 1). These mutants were analyzed in transient transfection assays in two cell lines, COS-7 cells and the neuroblastoma cell line SK-N-MC. Cells were expressed with a reporter plasmid carrying the luciferase gene driven by the GR-sensitive mouse mammary tumor virus (MMTV) promoter, a Renilla luciferase reference plasmid, and either an empty expression vector or a vector expressing one of the BAG-1M mutants. The light units measured for the reporter enzyme were normalized by the light units of the control enzyme (Renilla luciferase). Stimulation after incubation with 100 nM dexamethasone was in the range of 200 -300-fold and was set to 100%.
BAG-1M clearly inhibited GR-dependent reporter gene transcription, as expected (Fig. 1A). To our surprise, however, deletion of the E 2 X 4 motif domain had no influence on the ability of BAG-1 to inhibit GR-mediated transcription. On the other hand, deletion of the N-terminal 10 amino acids completely abolished the effect of BAG-1 on GR. The same pattern of activity of the BAG-1 mutants was observed in COS-7 cells and HeLa cells (data not shown). Because co-expression of various plasmids can produce misleading results in case of expression interference (49), we checked the relative amounts of GR and of BAG-1. The protein levels of GR were similar in all the exper-iments (Fig. 1B). Likewise, the levels of the expressed BAG-1 proteins were comparable throughout our experiments (Fig.  1C). We did not detect significant levels of endogenous BAG-1 in SK-N-MC or COS-7 cells. However, endogenous BAG-1 was detected in HeLa cells (data not shown). Thus, our results do not depend on the presence or absence of endogenous BAG-1. We conclude that the domain recently put forward as DNA binding domain of BAG-1 (39) is necessary for the inhibitory effect of BAG-1 on GR.
Spacing of the Lysine and Arginine Stretch Is Not Important for DNA Binding of BAG-1-To characterize the DNA binding domain of BAG-1 in more detail, we created a series of additional mutants (Table II); all proteins derived from BAG-1 mutants and isoforms in Tables I and II were expressed in bacteria and purified (Fig. 2, A and B). As a reference, we mutated lysines 2, 3, and 4 to alanines (BAG-1M KA), which has been described to abolish DNA binding of BAG-1. To reduce the DNA binding domain essentially to the positively charged amino acids lysine and arginine, we deleted amino acids 9 -67 (⌬9 -67). Moreover, we changed the spacing between lysines 2-4 and arginines 6 -8 by either deleting threonine 5 (BAG-1M ϪThr 5 ) or inserting one alanine (BAG-1M ϩAla 5 ) or two alanines (BAG-1M ϩ2Ala 5 ). After expression in bacteria, proteins were purified using Ni-NTA-agarose (Fig. 2, A and B, 1st lane of each mutant), and the histidine tails were cleaved with tobacco etch virus protease (Fig. 2, A and B, 2nd lane of each  mutant).
The purified and cleaved proteins were used to examine their ability to bind to DNA. The 125-bp fragment of an HindIII cleavage of DNA from bacteriophage was chosen as template. Gel shift assays (Fig. 2C) revealed that mutation of lysines 2-4 to alanines abolished DNA binding, consistent with a recent report (39). However, deletion of the E 2 X 4 domain (BAG-1M ⌬11-67; Fig. 2D) had no influence on DNA binding. Even additional deletion of two N-terminal amino acids retained DNA binding (BAG-1M ⌬9 -67; Fig. 2C). BAG-1S, as expected, did not bind to DNA. Therefore, it appears that the first 8 N-terminal amino acids are sufficient to confer the ability of BAG-1 to bind to DNA. Moreover, spacing between the positively charged amino acids lysines 2-4 and arginines 6 -8 is not important for DNA binding, because these mutants bind to   1. Effect of wild type BAG-1M, ⌬N10, and ⌬11-67 on GRdependent transcription. A, SK-N-MC cells were transiently transfected with 3.5 g of GR-responsive MMTV-luciferase (MTV-Luc) reporter gene, 1 g of an internal control plasmid encoding Renilla luciferase under the control of the SV40 promoter, and 4 g of a human GR-encoding plasmid (pRK7GRHA). In addition, 6 g of either an empty expression vector (C ϭ control) or the mutants BAG-1M ⌬N10, BAG-1M ⌬11-67, or the isoforms BAG-1M and BAG-1S were cotransfected. Cells were transfected using electroporation and replated and cultured for 24 h in fresh medium either with or without 100 nM dexamethasone. Luciferase activities were corrected by Renilla luciferase activities and are presented as percent activity with the activity of control vector-transfected cells set as 100%. Results represent mean values Ϯ S.E. of eight independent experiments performed in sextuplicate. B and C, representative Western blot of either GR or BAG protein.
Whole cell extracts used for luciferase assays were prepared for SDS-PAGE and immunoblotting. Antibodies were directed against either the C terminus of BAG-1 or the HA tag on GR.
DNA as efficiently as BAG-1M (Fig. 2, C and D). Therefore, BAG-1M apparently contains a short, unusual DNA binding domain.
Mutants That Inhibit DNA Binding of BAG-1 also Abolish Its Inhibitory Function on GR-The results shown in Fig. 1 suggest that DNA binding of BAG-1 is necessary for its inhibitory effect on GR function. We set out to either strengthen this correlation or to prove that, although this domain per se is necessary, it is not DNA binding but some other property of this domain that causes the inhibitory effect. Therefore, we constructed expression clones of all mutants and tested them in our transient reporter gene assay as described in Fig. 2. We find that BAG-1M ⌬9 -67, BAG-1M ϪThr 5 , BAG-1M ϩAla 5 , and BAG-1M ϩ2Ala 5 all are able to inhibit GR-dependent transcription (Fig. 3A). In contrast, BAG-1S and BAG-1M KA lost the inhibitory effect on GR. Again, the protein levels of GR were comparable throughout the experiments (Fig. 3B), as were the levels of the different BAG-1 mutants (Fig. 3C). Similar data were obtained in HeLa cells (data not shown).
Therefore, the data in Figs. 1-3 demonstrate that all mutants of BAG-1M that are able to bind to DNA inhibit the transcriptional activity of GR, whereas those that are unable to bind to DNA have no influence on GR activity.

The DNA Binding and the hsp70 Interaction Domain of BAG-1 Need to be Present in Cis to Inhibit GR Function-
Besides the N-terminal domain, deletion of the C-terminal 70 amino acids also abolishes the effect of BAG-1 on GR (40). Because these amino acids contain the interaction domain with hsp70, it has been proposed that interaction with hsp70 is necessary for the function of BAG-1. This raises the question whether the domains responsible for interaction with hsp70 and for binding to DNA are required to be present in cis. Therefore, we first used a point mutation of BAG-1 that abolishes interaction with hsp70 (R237A BAG-1M (42)). In our transient reporter gene assay, this mutant abolishes the inhibition of GR activity by BAG-1 (Fig. 4A, lane 4). This proves that the previously reported inability of a C-terminally truncated BAG-1 protein to inhibit GR is actually due to the inability to interact with hsp70 rather than to some other concomitant consequence of the deletion. To test whether a BAG-1 protein that is unable to interact with hsp70, but has an intact DNA binding domain, can rescue the inhibitory function of a DNA-binding mutant bearing an intact hsp70 interaction domain, we concomitantly expressed a DNA-binding mutant (KA) with an hsp70 interaction mutant in our reporter assay. Although these proteins are expressed to a comparable level as wild type BAG-1 (Fig. 4B), no inhibition of GR-dependent reporter gene transcription was detected (Fig. 4A, lanes 5 and 6). We also show that the hsp70 interaction mutant does not prevent inhibition of GR by BAG-1M (Fig. 4A, lanes 7 and 8). The levels of GR were the same throughout our experimental conditions (Fig. 4C). We conclude that the N-terminal domain and the hsp70 interaction domain of BAG-1 need to be present at the same time and on the same molecule. DISCUSSION BAG-1 displays a remarkable functional versatility by participating in regulation of apoptosis via interaction with Bcl-2, by functioning as a modulator of chaperone activity via interaction with hsp70, by enhancing transcription in general (50), by regulating neuronal differentiation (3), by mediating stress signaling via regulating Raf-1/ERK (51), and by modulating nuclear receptor-dependent transcription (shown here and see  2. A and B, bacterially expressed BAG-1 and mutants. BAG-1 isoforms and mutants were cloned into the bacterial expression vector pProexHTa (Invitrogen) that delivered a histidine tail to the N terminus of the proteins. BAG-1M isoforms and mutants were expressed in E. coli, purified using Ni-NTA-agarose columns following the QIAexpressionist protocol (Qiagen Inc., Hilden, Germany). The histidine tail was removed by cleaving with tobacco edge virus protease, and proteins were analyzed by SDS-PAGE. Staining was performed with Coomassie Blue R-250. Noncleaved and cleaved proteins were loaded alternately onto gels. Refs. 9, 12, 13, and 40). This functional diversity suggests that BAG-1 uses different domains for different functions. For example, it has been reported recently (52) that distinct BAG-1 isoforms have different anti-apoptotic functions. In the present study, we focus on the requirements for BAG-1 to inhibit GRdependent transcription.
It has been speculated that the serine-and threonine-rich E 2 X 4 domain is necessary for the inhibition of GR function by BAG-1 (38). However, we demonstrate that deletion of the entire E 2 X 4 motif domain does not affect the ability of BAG-1 to counteract GR-dependent transcription. This explains the recent finding that mutations of phosphorylation sites within this domain retain the activity of BAG-1 (40). Against expectation, we discovered that a small DNA binding domain of only 8 amino acids is necessary for inhibition of GR function. This DNA binding domain is uncommon for transcriptional regulatory proteins, because DNA binding appears to be nonspecific (39). The nonspecificity of DNA binding is in line with our observation that spacing of the two positively charged stretches of three lysines and three arginines is not important for DNA binding. This suggests an electrostatic interaction of these positive charges with the negatively charged phosphate backbone of DNA.
Another domain of BAG-1 essential for inhibition of GR-dependent transcription is the hsp70 interaction domain (38). By employing a point mutation that interferes with binding to hsp70, we were able to prove that the importance of this domain is not merely structural stabilization of the protein but that interaction with hsp70 indeed is necessary. Moreover, we demonstrate that binding to hsp70 and binding to DNA have to be linked on the same molecule, i.e. a mixture of two BAG-1 proteins that either carried a mutation in the one or the other domain showed no effect on GR-dependent transcription.
What are the structural requirements of BAG-1 for its effect on other nuclear receptors? In the case of the androgen receptor BAG-1 enhances transcription, but the domains required are similar at first sight, because the N-terminal domain of the long isoform BAG-1L and the hsp70 interaction domain are both required (9). However, it appears that the N-terminal domain of BAG-1L is necessary only for nuclear targeting, because BAG-1M is inefficient unless it is forced to the nucleus by adding a targeting sequence (9). Similarly, the vitamin D receptor also has been reported to be enhanced by BAG-1L but not by BAG-1M. Moreover, the hsp70 interaction domain is required (10). It should be noted however, that more recently an inhibition of vitamin D receptor by BAG-1L has been reported (11). In the case of the retinoic acid receptor, an inhibition by BAG-1M has been demonstrated (12), but no domain analysis was conducted so far.
Thus, taken these reports on the effects of BAG-1 on nuclear receptors together, it seems clear that, although the effect of BAG-1 can be either inhibitory or stimulatory, the N-terminal and C-terminal domains are required. We could prove that the function of the C terminus is indeed the interaction with hsp70. The N terminus of BAG-1L has been suggested to be important for nuclear targeting of BAG-1. However, GR and the retinoic acid receptor are inhibited also by BAG-1M. Moreover, BAG-1M can be translocated into the nucleus upon activation of GR by hormone (40). Thus, we propose that the additional N-terminal amino acids of BAG-1L are necessary only for those receptors that do not promote nuclear transport of BAG-1M. This is also supported by the observation that BAG-1M is effective on vitamin D receptor upon forced nuclear targeting (9). We discovered that the N-terminal 8 amino acids of BAG-1M, which probably represent the entire DNA binding domain, are also required for its function on GR. It will be interesting to evaluate the function of this domain for the effect of BAG-1M on other nuclear receptors.
Regarding the function of the DNA binding domain of BAG-1M, it is of note that BAG-1M binds nonsequence-specifically to DNA. Its intracellular concentration can be estimated from careful analyses in a variety of tumor cell lines (30,35) to be one molecule per 10 -100 kb of DNA. Moreover, it is a predominantly cytosolic protein (30,40). Therefore, other factors must determine when and where BAG-1M binds to DNA. We propose the following model: BAG-1 is transported into the nucleus along with GR upon activation of the receptor with hormone (40). GR interacts with the chromatin at glucocorticoid response elements. The first step in GR-mediated transactivation presumably is remodeling of the local chromatin structure by cofactors recruited by GR (53). Once the chromatin is restructured, BAG-1 gains access to the DNA, which interferes with the further functions of GR for efficient transactivation. Access of BAG-1 to the DNA might be promoted by the rapid exchange with regulatory sites of GR (54).
The guidance of a nonsequence-specific DNA-binding protein to specific chromosomal loci as suggested here for BAG-1 is not without precedence; Cdc6 is an essential protein in yeast which is recruited to replication origins in the G 1 cell cycle phase by another replication factor, origin recognition complex protein 1 (55). It is intriguing not only that cdc6 alone, like BAG-1, binds nonspecifically to DNA but also that site-directed mutagenesis identified the basic protein motif KRKK as essential for DNA binding and function of the protein (56). Apparently, this motif  7 and 8). Cells were transfected using electroporation, replated, and cultured for 24 h in fresh medium either with or without 100 nM dexamethasone. Luciferase activities were corrected by ␤-galactosidase activities and are presented as percent activity with the firefly luciferase activity of control vector-transfected cells set as 100%. Results represent mean values Ϯ S.E. of four independent experiments performed in sextuplicate. B and C, representative Western blot of either GR or BAG protein. Whole cell extracts used for luciferase assays were prepared for SDS-PAGE and immunoblotting. Antibodies were directed against either the C terminus of BAG or the HA tag on GR.
is very similar to the basic motif of the N terminus of BAG-1 that we demonstrate is not only essential for binding to DNA but also for functional integrity with respect to the inhibition of GR-dependent transcription.
Finally, because we demonstrate that DNA binding and interaction with hsp70 are required in cis, it is likely that hsp70 is required during this process, i.e. hsp70 might be present during chromatin remodeling or even promoting it. In general, chaperones like hsp70 and hsp90 are considered to act in the cytosol. However, nuclear translocation of hsp70 along with hormone-activated GR and BAG-1 has been reported (40). In addition, there is ample evidence that hsp90 is involved in subcellular trafficking and chromatin recycling of GR (57,58), expanding the original observation that the Hsp90/Hsp70based chaperone machinery of reticulocyte lysate could dissociate the hormone free GR from DNA and regenerate the nonDNA-binding GR-Hsp90 heterocomplex (59). Moreover, a nuclear role in regulating the chromatin assembly and activity of nuclear receptors has been reported recently for the hsp90interacting chaperone p23 (60,61). An open question is whether the described disassembly of transcriptional regulatory complexes by p23 requires ATP (61), as the requirement of hsp90 or hsp70 for the effect of p23 remains to be elucidated.
In summary, overwhelming evidence has accumulated that molecular chaperones and associated proteins like BAG-1 have distinct functions in the nucleus. Obviously, DNA binding of BAG-1 is a clear indication of a nuclear function of this protein.
Our model proposed above also implies that BAG-1 may contribute to promote disassembly of transcriptional regulatory complexes by an Hsp70-based process, similarly to p23, which may act via hsp90. This model is also consistent with the observation that DNA-independent transrepression by GR is not inhibited by BAG-1 (38). Future experiments will show whether the DNA binding domain is important for other functions of BAG-1 and whether the model proposed here can also be applied to other nuclear receptors.